Principles of Data Science. Tiny Data Science Project

Elnara Mammadova (210026788)

1. Dataset, Background

https://www.equinor.com/en/what-we-do/digitalisation-in-our-dna/volve-field-data-village-download.html

For the first time, all subsurface and production data from a field on the Norwegian continental shelf - VOLVE field was disclosed by Equinor. Files were acquired thourgh Azure where all the dataset is stored.

The Equinor HRS Terms and conditions for licence to data Volve

We will produce petrophysical analysis of three wells from Volve field: 15/9-F-11 T2, 15/9-F-11 A, 15/9-F-11 B and provide facies classification based on the derived porosities and permeabilities.

2. Wells' Information

15/9-F-11 will not be analysed, as the well path did not achieve sufficient separation from nearby wells, hence P&A before a technical side-track was performed for 15/9-F-11 T2.

2.1. 15/9-F-11 T2 - Appraisal, Abandoned

Objective of the 15/9-F-11 pilot well was to prove oil in the Volve North prospect segment and to define the field OWC. In addition the pilot was placed such that it could also prove a shallower Hugin interpretation based on the previous 2009 interpretation, which represents an additional prospect upside.

The technical side-track, 15/9-F-11 T2, was kicked off from F-11 @ 257 m MD RKB. TD @ 4562 m MD RKB. Oil show is described on cuttings @ 4411 m MD.

These 2 residual oil pockets are observed in the following intervals:

LWD log data is of good quality. No oil and water samples were taken since no commercial find.

2.2. 15/9-F-11 A - Appraisal, Producing Oil

Objective of the second pilot 15/9-F-11 A was to prove oil and sufficient reservoir thickness in the Northwest fault segment.

15/9-F-11 A was side-tracked from F-11 T2 with KOP @ 2586 m MD RKB. TD @ 3762 m MD RKB. Oil Down To (ODT) ~ 3702 m MD RKB / 3025.5 m TVD MSL which corresponds to Top Sleipner Fm.

LWD log data is of good quality. No water sample was taken since water was not encountered below the oil column.

2.3. 15/9-F-11 B - Infill, Producing Oil Well

Since 15/9-F-11 / F-11 T2 found no or insufficient oil in the Volve N prospect, the objectives of 15/9-F-11 B was to be side-tracked to an infill location between existing producers F-12 and F-14 in the main field and downflank towards existing injector F-5. The aim was then to geosteer the well in the upper part of the Hugin reservoir, which was expected to be only partially drained.

15/9-F-11 B was side-tracked from F-11 T2 with KOP @ 2585 m MD RKB. TD @ 4770 m MD RKB. Hugin Fm. was oil filled in large parts of the well, and encountered also oil in the first part (shallowest) of Hugin Fm. Below ~ 4701 m MD RKB / 3148 m TVD MSL (a coal layer) the water seems to be initial, due to the low resistivity.

LWD log data is of good quality. 8 intervals have been perforated.

Information acquired from Equinor Petrophysical Report, (November 2013)

3. Import Libraries

4. Load Wells

4.1 Quicklook Analysis

4.2. Load Well Las Files

4.3. Well Headers

4.4. QC

 4.5. Convert Well objects to DataFrames

 4.6. Formation Tops (reservoir section)

F_11T2

F_11A

F_11B

Well F_11B doesn't have Sonic logs (DT, DTC)

 5. Vertical Composite - Raw curves

5.1. F_11T2 Plot
5.2. F_11A Plot
5.3. F_11B Plot

6. Data Preparation

6.1. Outlier and noise Detection - visual analysis

6.1.1. MSNO plots

Missing data are towards the leg of the legs, since the measurements are spaced along the tool length. We can go ahead and remove all the nan values.

 6.1.2. BoxPlots

6.1.3. Label Formation Tops

6.1.4. DistPlots

6.2. Remove noise

We will apply filter for noise removal, in an attempt to remove outlier noise while preserving the relative magnitude of the curves.

Check for any remaining outliers/noise

6.3. IsolationForest Outlier Removal - F_11B

Since we will be using RT in our Water Saturation estimates, they need to be as clean as we can get them. Resistivity logs in Hugin formation in F_11B still show some outliers after noise removal, due to the affects of polarization horns. We will be removing these outliers using IsolationForest.

6.4. Smooth the Curves

6.5. DTC prediction for 15/9-F-11B

7. Load DLIS files for Mudlog Lithology

https://towardsdatascience.com/loading-well-log-data-from-dlis-using-python-9d48df9a23e2

7.1. View Headers

7.2. Review Frames

7.3. Create Channels

7.4. Pull Lithology and Depth Curves

7.5. Create Mudlog DataFrames

 7.6. Merge mudlog lithology with las Dataframes

8. Visualizing Clean Data

8.1. F_11T2 Plot

8.2. F_11A Plot

8.3. F_11B Plot

 9. Formation Evaluation

9.1. Evaulation Parameters

Below are the parameters we will be using to evaluate the Porosity, Permeability and Saturation of the formations

As described in petrophysical report: “Sleipner Øst and Volve Model 2006, Hugin and Skagerrak Formation, Petrophysical Evaluation”. November 2006. Author: Elin Solfjell, Karl Audun Lehne.

9.2. Total Porosity

Total porosity, PHIF (φF ), is derived from the density log which is calibrated to overburden corrected core porosity for wells drilled with either OBM or WBM.

The Neutron log, NPHI, has been used to correct for varying mud filtrate invasion.

where:

F_11T2

F_11A

F_11B

9.3. Shale volume

To determine VSH, the standard model VSHGR from linear GR relationship is applied:

VSH = (GR − GRmin)/( GRmax −GRmin)

where:

F_11T2

F_11A

F_11B

9.4. Permeability

Calculate Permeability The horizontal log permeability, KLOGH, is derived from the following equation based on multivariable regression analysis between log porosity and shale volume (“normalized” against VSHDN) against overburden corrected core permeability:

F_11T2

F_11A

F_11B

 9.5. Water Saturation

Water saturation is calculated using Archie equation, giving a total water saturation:

Swt = [(a×Rw)/((φFm)*Rt)](1/n)

where:

Calculate cementation exponent

Water resistivity of formation at formation temperature -- RW

Rw(TRW) - Rw measured at temperature TRW - temperature at which Resistivity was measured FT - Formation Temperature KT1 = 21.5 (for Metric units)

F_11T2

F_11A

F_11B

9.6. Bulk Volume of Water

9.7. Sand Flag

9.8. Evaluation Logs

9.8.1. F_11T2 Plot

9.8.2. F_11A Plot

9.8.3. F_11B Plot

9.9. HUGIN Formation summary

10. Facies Classification

We will be using VSH, RHOB, NPHI, PHIT, KLOGH logs for facies classification, as they are independent of reservoir fluid saturations.

10.1. Final Composite Well

 10.1.1. F_11T2

 10.1.1. F_11T2

 10.1.3. F_11B